DLBCL - Episode 3

Zelenetz Discusses How to Evaluate Responses to Therapy in DLBCL

Andrew Zelenetz, MD, PhD, medical oncologist at Memorial Sloan Kettering Cancer Center, is the current chair of the Non-Hodgkin Lymphoma Guidelines panel of the National Comprehensive Cancer Network (NCCN).

Andrew Zelenetz, MD, PhD, medical oncologist at Memorial Sloan Kettering Cancer Center, is the former chief of the Lymphoma Service and the current medical director of Quality Informatics. He has helped develop several agents now approved to treat lymphoma, including 131I-tosituomab/tositumomab, bortezomib and pralatrexate. Zelenetz is the current chair of the Non-Hodgkin Lymphoma Guidelines panel of the National Comprehensive Cancer Network (NCCN) and vice chair of the Lymphoma Core Committee of the Cancer and Leukemia Group B. The author of more than 100 papers on lymphoma research, Zelenetz studies both novel agents and the prognostic value of patients’ pathology specimens, using computer-aided image analysis. He spoke with The American Journal of Managed Care® (AJMC®) about unmet need in the treatment of diffuse large B-cell lymphoma (DLBCL), how to evaluate responses to therapy, and possible ways to identify which patients should be early candidates for chimeric antigen receptor (CAR) T-cell therapy, perhaps with next-generation sequencing (NGS). This interview has been edited lightly for clarity.

AJMC®: What do you consider to be the largest unmet needs in diffuse large B-cell lymphoma, or DLBCL?

Zelenetz: DLBCL is a heterogeneous disease. We can demonstrate it is made up of at least 7 molecular subvariants. And we've made great progress with treatment—60% to 65% of patients treated with rituximab and CHOP chemotherapy can be cured. But there are many patients who are destined to progress after treatment. The recent trial of pertuzumab added to rituximab, cyclophosphamide, doxorubicin, and prednisone vs CHOP demonstrated improvement in progression-free survival, but thus far, no improvement in overall survival. So, we obviously need to make progress in the first line.

One thing is, these molecular subtypes we've identified are really laboratory-based things. We don't have the clinically applicable tools to turn this around quickly, so that we can test targeted treatments that would be appropriate for these different subtypes. That’s one of the areas in the first line, once the disease recurs, where we have lots of unmet need—because the vast majority of patients with relapsed and refractory disease simply aren’t cured. Again, we've made great strides—the most standard treatment that we've used for many years in the relapsed setting is high-dose therapy with autologous stem cell rescue, sometimes co-autologous stem cell transplant. This is a curative regimen, but it's a curative regimen based on intent to treat in about a third of patients, and patients who have short remissions of less than a year or simply don't go into remission—they actually have very, very poor outcomes….

Now recently, 2 prospective randomized trials have shown CAR T-cell [therapies] are potentially an improvement in that space. And axicabtagene ciloleucel (Yescarta) was recently approved in that space. There was a third trial with tisagenlecleucel (Kymriah) that was negative. So not everything is showing exactly the same results. And in the third line, CAR T cells have emerged in important treatment, but this itself represents an unmet need. CAR T cells are not universally available. They're not universally appropriate for all patients because of the side effect profile. And many patients never get to CAR T cell because their disease biology or because they lack access to the drug. So, there are lots of unmet needs here in the relapsed space in large cell lymphoma.

AJMC®: You play a critical role as a member of the NCCN B-cell lymphomas guidelines panel. Can you walk us through your process of digesting and evaluating new information? What is your evaluation process, and what is the role of something like real-world evidence as you go through these regular guideline updates?

Zelenetz: The guidelines are updated in a 2-day meeting once a year, and then will be updated as needed other times during the year, particularly if there are new approvals. We frequently have a brief meeting after the annual meeting of the American Society of Hematology. We have that in January to make sure that there are no presentations or data that would require changes to our guidelines. This is ongoing process. As chair, I go through what I think are the most important data of the year, I ask members of the panel to prepare presentations to review those data so that everyone is up to date. They know what the data are. And then we discuss whether these changes and this new information should be reflected in changes in the guidelines. Sometimes the answer is no, it's not a significant enough change, or the other times, yes, the standard of care has changed. And new treatments are added to the guidelines.

We use data from prospective randomized trials. We use data from large, well-done phase 2 studies. We prefer published full manuscripts, but sometimes full manuscripts can be a year or year and a half after critical data has been presented. So [we try to] get the data ourselves to review—and most investigators are very generous and will provide the NCCN panel with the data, so that we can review it. But real-world data is a legitimate source of data. Now, again, just like everything else, we're going to look at the quality, the consistency, are the data appropriate to use to make guideline recommendations. But if we feel they are, then there's no reason why we would say, real-world data shouldn't be used.

The NCCN guidelines are a little different than ASCO guidelines. ASCO guidelines do a really deep dive in a particular area and make a decision. ASH guidelines are very similar, but the NCCN guidelines are continuum of care from diagnosis through relapse, through second treatment, through third treatment, through fourth. And we don't have prospective randomized trials for every single point along a treatment decision. So sometimes we have to use the data that's available to us. And in certain circumstances that's going to include real-world data to help inform us as to what we should do, so we can make recommendations across the continuum of care.

AJMC®: How do you evaluate responses to therapy in DLBCL? What are you looking for to determine whether a patient is successful in the first line, whether they're potentially successful in the second line or are they heading to transplant and you're going to make sure they're on appropriate therapy to go down that route, or perhaps to CAR T therapy, for example?

Zelenetz: Well, the response evaluation ironically starts before we start treatments; we have to make sure we know where the disease is. Pretty universally today FDG PET (fluorodeoxyglucose-positron emission tomography) is used to identify sites of disease in large cell lymphoma. More than 98% of patients with DLBCL have FDG avid disease. The bone marrow biopsy is often recommended, but if a PET scan clearly shows bone disease, there's absolutely no reason to do a bone marrow biopsy in that setting, because you already know that the patient has involvement of the bone. Sometimes other sites need to be evaluated, and the brain might be appropriate in certain circumstances; lumbar puncture with flow cytometry might be appropriate. That’s all done before, so we know where the disease is. Then after treatment, we want to look where the disease was originally—is it all gone?

Now, we do have a number of different response criteria. We have the Lugano response criteria, which look at both size of lymph nodes, as well as guidance for use of FDG PET scan. In the Lugano response, we look at the sum of the perpendicular diameter. So, we measure lymph nodes in 2 dimensions, multiply those 2 numbers and add up to 6 indicator lymph nodes. And we also look at the PET scan; the PET scan is rated based on how much uptake remains relative to normal organ. So, if there's no uptake whatsoever, it's a 1; less than mediastinal blood pool, it’s 2; less than liver, it's a 3; more than liver, but not a lot it’s four; and a lot, or new lesions are 5. You can see that 4 and 5 are a little bit hazy—and it is a little bit hazy. But typically, in most studies Lugano 1, 2, and 3 is determined as a negative, or a complete response and a Lugano 4 or 5.

Now, in studies where you're trying to limit the amount of treatment and you want to say, “Can I use the minimum amount of treatment?” Then sometimes a more stringent criteria can be used, where Lugano 1 and 2 would be a complete response. And 3, 4, and 5 would be considered incomplete responses. The RECIL criteria also use PET, and it's actually in a very similar way. But the big difference is that when lymph nodes are measured, they're measured in unidimensional and it's up to 3 lymph nodes. There was some debate whether we should use the short-axis dimension or the long-axis dimension; we used long-axis dimension in RECIL, and by comparing studies that were done with RECIL and originally done with Lugano the results are actually very comparable. And so we have reproducible ways of assessing response.

Now does everyone use a Lugano response criteria for their patient in clinic? Most don't. Those are really were designed for clinical trials. The more academic, the more research-oriented, the more likely you're going to see in someone has a Deauville 3 response, or Deauville 2 response or a Lugano 2 or 3 response.

AJMC®: It appears clinicians use PET scans to gain as much information as possible about a patient, and then as you monitor their response to therapy. Along those lines, I know you recently published a phase 2 study in which you used next-generation sequencing, or NGS, to look at the dynamics of therapy in particularly CLL (chronic lymphocytic leukemia). How might you see NGS monitoring help build that repertoire of information, similar to what you have with a PET scan, as you treat relapsed/refractory DLBCL in the future?

Zelenetz: Well, so one of the limitations of PET scans is if we take all comers, there are more relapses in people who have negative PET scans than in people who have positive pet scans. We know that a PET scan isn't adequate to identify what we call minimal residual disease; it’s simply not sensitive enough. Recently in a CLL study we looked at next-generation sequencing for immunoglobulin gene rearrangements. And we actually found that the kinetics of response, or how quickly someone responded, divided patients into 2 groups, those that achieved undetectable minimal residual disease rapidly and actually were able to have a short duration of treatment, and a second group that took longer to achieve that same undetectable minimal residual disease; this group up getting substantially longer treatment. Similar data were generated by the Alizadeh Laboratory at Stanford. They looked at cell-free DNA and found that if you look very early after the initiation of workshop chemotherapy, if the level of ctDNA dropped by 2 logs, that was a very good predictor for outcome. Now, of course, these are just dividing patients biologically, using basically a real time method to identify people with favorable or unfavorable biology. But in large cell example, if someone doesn't achieve that rapid response, then you could conceive of taking them to a second-line treatment, maybe CAR T cell therapy or something else because you know that group is going to have a very poor long-term outcome.

AJMC®: Speaking of similar population of patients, we know many patients are unable to proceed to transplant after their first-line therapy for a variety of reasons. What factors do you consider when you're selecting that DLBCL patient for second-line therapy who wouldn't be a candidate for transplant or can't make it to that stage in therapy?

Zelenetz: So eligibility for high-dose therapy and autologous stem cell rescue is extremely variable. Essentially, there are some centers where there's literally no upper limit in the assessment of fitness; other centers will say, if you are 70 years and a day, you're not eligible for transplant. We don't have great hard and fast rules for this. But everyone is going to have someone they won't transplant and sometimes their comorbidities will prevent anything relatively intensive, but it turns out that CAR T cell therapies like tisagenlecleucel or lisocabtagene maraleucel are much better tolerated than high-dose therapy and autologous stem cell rescue, less demanding, able to be given to people with comorbidities. So that represents a potentially curative option for that group of patients who are “transplant-ineligible,” but then there are those who are really not eligible for any cellular therapy, anything aggressive, and those are the patients who we really want to put on clinical trials where we're looking at really novel new treatments that could address critical pathways that might be abnormal in that patient's tumor with a more targeted approach that might have less toxicity.

AJMC®: Do you think something like NGS will help us better identify those patients and as we develop more and more targeted therapies? Could we use it across the spectrum in DLBCL, where NGS might be able to help us find out early on which patients would be eligible for such a therapy?

Zelenetz: Well, it comes back to what I said earlier: do we have the right tools to be able to translate molecular subtypes from the laboratory, which we can apply retrospectively to patient populations? Can we apply that prospectively and identify patients? Next-generation sequencing is an important component of that. And we can turn this around—using cell-free DNA will frequently help us with that. And that's one of the things that the Alizadeh approach showed, because they used cell-free DNA, but looked at a panel of genes. So, you could see what genes were mutated in the resistant population. And if there were appropriate targeted treatments for some of those mutations, you could have potentially have a clinical trial for that group. But yes, we need to be able to better characterize the molecular subgroups so that we're not just throwing new treatments at people, but at least there's some information. For instance, if you're an activated B cell, you're more likely to have activation of NFκB downstream, which is downstream of the CARD11-BCL10-MALT1 complex, the CDM complex. There are novel inhibitors that can target MALT1, and early studies have shown some promise here.

Part of the problem that we have in large cell lymphoma is we've identified a number of mutations and genes, but those genes don't appear to be druggable. A perfect example is BCL6. BCL6 is a critical transcription regulator, but it just doesn't seem to be druggable because the way it works is through protein interaction. But there's a new approach with proteolysis targeted chimeras, or PROTACS, where we can promote the degradation of these proteins and then potentially target them by making them disappear. And so, we have lots of potential options including new antibodies both naked antibodies, as well as antibody-drug conjugates. And then there's the important area that is appropriate for some people who can't get CAR T cells, but could get what we call bispecific antibodies. And these are antibodies in which one part of the antibody binds to a B cell, often CD20; it's been a pretty common target. On the other hand, the target can be a T-cell target and by bringing these two together, you activate T cells and try to get them to kill the tumor. So bispecifics, are another important potential option.

AJMC®: What do you consider to be the most promising developments as you look forward in relapsed and refractory DLBCL?

Zelenetz: Well, as I said, I think the bispecific antibodies are critical here. Novel antibody-drug conjugates combinations of new naked antibodies are important. But I actually also think that attacking some of these undruggable targets with PROTACS is going to be important better targeting of the clearly abnormal pathways such as with the MALT1 inhibitors or further upstream with BTK inhibitors are potential opportunities for certain mutations like EZH2 mutations. We have EZH2 inhibitors that have activity in a large cell lymphoma. So, I think it's going to be really understanding the biology of the disease and trying to get patients with a more reasonable likelihood of responding, not put every patient in those same bucket, but trying to [00:24:00] select patients based on their biology.

AJMC®: Is there anything else you would like to share with our audience?

Zelenetz: DLBCL represents an oncology success story. In first-line, we are curing 60% to 65% of patients. If you're treating a lot of lung cancers, it looks like the lymphoma doctors have it easy. But we've made relatively limited progress since the early 2000s, when we added rituximab to CHOP, despite our vast understanding of the biology. And this is in part because of that difficulty in translating those scientific discoveries into the clinic. In large part, because it's been hard in real-time, in the appropriate way to identify which lymphomas a patient has so that we can test different treatments for patients. We're making some headway there—and we're optimistic that we can do that. But even without these new drugs, there are opportunities I think to improve patient care and outcomes at major medical centers adjusted for patient mix, because that's always the big criticism. But when you adjust for patient mix it’s still better than in the community. Now this is part related to accuracy of diagnosis, management of treatment-emergent side effects, but most importantly, I think particularly in the relapsed/refractory setting access to new and innovative therapies, as this is often limited in the community. It's limited by payers as well as Medicare, and so increasing access to approved and available novel treatments can improve outcome and getting more patients from the community onto clinical trials to test some of the new hypotheses is going to be really important.